Election 2016: The big data trail to our next president
You better watch out,
You better not cry,
Better not pout,
I'm telling you why:
Big data is coming to town.
According to Wired, in 2012, "Obama's campaign began the election year confident it knew the name of every one of the 69,456,897 Americans whose votes had put him in the White House."
Let's be clear. This is not about the NSA or the FBI or the CIA. The president's election team wasn't able to push right through your apparent privacy using police state tactics.
Oh, no. They used something far more dangerous and intrusive: Math.
We are all aware that we leave a digital footprint everywhere we go, but the level of detail in that footprint can be breathtaking.
What you buy, what you read, what you share online, who you associate with, what your mood is, where you work, what you do, what your health situation is, where you've donated, what clothing styles you like, what car models you buy, your favorite Cola brand, your favorite phone brand -- all of that information is available to those with the budget to buy it and the algorithms to aggregate and sift through it.
This is where big data is changing the face of American election politics. For most of the 20th century, election politics was often about putting voters into demographic boxes and then reaching out to those boxes.
In other words, you were a soccer mom, an up-and-comer, a broke-and-angry, a comfortable middle-ager, etc. Each of those demographic and psychographic categories was considered enough to help a campaign target resources. All they had to do was count up who fit into what category, overlay that to a region, and they had their plan.
But what happens if you have 18 percent staunch conservatives living in the middle of a broadly liberal community? In the past, those outliers were ignored by retail politics, and the only way they could be reached was through national campaigns, debates, and ads.
In 2008 and 2012, that all began to change. Now, rather than targeting broad groups of voters, savvy campaigns are targeting you, the individual voter. Each individual over 18 years of age has a very complex and very lengthy dossier available to the campaigns willing to embrace data science.
Analysts can predict who you are going to vote for -- often before you, yourself, have made up your mind.
I spoke recently about this with Dave Wakeman of the Wakeman Consulting Group. Dave has some history in this business. He worked on the 2012 campaign, and wrote ads in support of President Obama for the AFL-CIO, AFSCME, and others. He also worked with MoveOn, the International Association of Fire Fighters, and the Maryland Fraternal Order of Police President.
According to Wakeman: "In 2012, the Obama team used data to target specific voters and make many more targeted conversations. One way that they tailored their messages through the use of data was in their fundraising appeals in email ... depending on your profile and the data they had associated with your email and profile, you would receive a different email with a different ask."
Of course, not all was perfect. Wakeman reported, "So if you had more than one email address, you might see asks for different amounts of money, different actions, and different messaging."
Even in the largest campaigns, there is clearly room for improvement.
As we move forward into 2016, big data will take on an even greater level of importance, and we're moving into a world where real-time big data becomes not only possible, but also economically practical.
This is because of the tremendous decline of RAM cost, which makes in-memory computing possible for large data sets. RAM is millions of times faster than hard drives, making analysis tasks that might take weeks to perform using spinning platters generate results within seconds.
But big data in presidential politics isn't about know whether you've been bad or good; it's about resource allocation. And this is where big data becomes game changing.
Think for a minute about the nuts and bolts of a presidential election. Where do you spend your ad dollars? What do you say? What communities do you target? Where do you set up campaign workers? Your headquarters? What doors do you tell volunteers to knock on? What do you tell them to say? What phones do you ring? What do you ask? Who is likely to donate? What is likely to sway big donors?
How much? Where? What? When?
This is where data analytics takes the front seat. The more rapidly you can move your resources, the more responsiveness your campaign can be to shifts in mood and counter-tactics by the competition.
Wakeman spoke about that in relation to the 2012 Romney campaign:
The Romney campaign never really embraced data in a way that allowed them to make smart decisions. A really good example is in polling. Somehow, the internal polling the Romney campaign was using to make decisions was counter to every other publicly available poll and any internal polls that the Democrats had. This made for decisions like Romney campaigning in Pennsylvania on the closing weekend of the campaign.
In other words, the Romney campaign allocated the wrong resources to the wrong state at the wrong time. And lost.
But there is more to consider than just resource allocation. There will be a lot of very sensitive data flowing through the campaigns. I discussed this concern with Joel S Winston, former deputy attorney general in New Jersey. Joel is also former director of Digital Media for a prominent DC political consulting firm, and now chairs the cybersecurity law practice at Winston Law Firm, LLC, in New York City.
He cautioned that, "The theme in 2016 will be privacy of voter data. Campaign databases will contain highly personal and sensitive data on hundreds of millions of American voters."
He warned that a data breach by a presidential campaign would pose significant risk.
Winston made some additional points worthy of concern:
Political campaigns and entities that collect, analyze, and act upon personal voter data have multiple legal responsibilities to protect the data and limit access. Modern campaigns have an enormous task to protect the big data they give to staff, vendors, and campaign volunteers.
He shares some of my concerns about security. "Every presidential campaign with an electoral strategy for victory must also have a rock-solid privacy plan to protect operational and big data files."
Finally, Winston made a simple statement that could be a real wake-up call to campaign operators everywhere: "An epic data breach can sink a modern campaign."
In looking toward 2016, Wakeman made a strong point: "Each side should be using data as a way to encourage meaningful interactions with their voters. The challenge for any campaign is going to come in the form of having access to reams of data and using it in a meaningful way. In campaigns, the challenge is truly finding ways to make the data actionable and not just theoretical."
That challenge applies not just to politics, but all data science. Having the data is one thing. Using it as a force for good is another thing entirely.
They see you when you're sleeping.
They know when you're awake.
They know if you've been bad or good,
So be good for goodness' sake!
PS I am already getting calls from the Hillary Clinton campaign, and it's barely 48 hours after her announcement. Here's a useful data point for anyone on her team: The more you call, the more likely I am to vote for anyone else.
By the way, I'm doing more updates on Twitter and Facebook than ever before. Be sure to follow me on Twitter at @DavidGewirtz and on Facebook at Facebook.com/DavidGewirtz.